Conceptualizing and measuring motivations for music use

Scientific endeavors to describe the potential functions, benefits, or uses of music can take several forms. While some researchers have addressed this theoretically, such as seeking to explicate the evolutionary origins of music [26,27], others have sidestepped evolutionary claims and approached this empirically, focusing on the more tractable patterns of everyday music use, like looking at individual’s listening preferences and self-reported reasons for music listening [e.g., 2,15]. Combined, the two approaches have generated a particularly abundant – and somewhat overlapping – list of potential functions. However, as pointed out in a literature review by Schäfer and colleagues [1], while the number of distinct dimensions have varied across studies and theoretical frameworks, the existing literature could broadly be distilled down to a few functions, each associated with cognitive or self-related processes (e.g., escapism, intellectual, or eudaimonic experiences), emotional purposes (e.g., to increase positive emotions), social processes (e.g., maintenance of interpersonal relationships), as well as physiological processes (e.g., reduce/enhance arousal). Schäfer and colleagues conducted an empirical investigation of 129 non-redundant functions, suggesting three distinct dimensions: achieving self-awareness, regulating mood and arousal, and expressing social relatedness.

To measure interindividual variability in music use, several self-report questionnaires have been developed, most of which broadly probe the above-mentioned basic dimensions of musical functions. One key example is the 30-item scale that we utilized in the present context: the Music use Motivation (MM) module of the Music Use and Background Questionnaire (MUSEBAQ) [25], which comprises the four dimensions of musical transcendence, emotion regulation, social bonding, and musical identity and expression. The transcendence subscale contains items that tap into eudaimonic, cognitive, and creative functions, in addition to strong sensations and transcendental experiences. The emotion regulation subscale captures the tendency to use music-related strategies to regulate or alleviate (negative) emotions, for example, through diversion, discharge, or solace. The social bonding subscale probes motivations related to social-relational regulatory strategies, and lastly, items of the identity subscale relate to the expression of identity through music.

Another relevant self-report scale is the Uses of Music Inventory (UMI) [15], which contains 15 items categorizing music use into the three main factors of emotional regulation (e.g., to induce, alter, or maintain emotional states), background use (e.g., listening to music while working), and cognitive use (e.g., using music for intellectual stimulation). Although there is some overlap among the UMI and MM subscales (e.g., both include an ‘emotion regulation’ dimension), a notable strength of the MM is that it expands the phenotypic spectrum to include dimensions like musical transcendence, which occupies a central position in some theoretical models of musical functions [e.g., 28] and aligns with qualitative accounts describing strong experiences with music [29].

But why do people differ in their preferred ways of engaging with music? Investigating the origins and nature of individual differences in music use motivations may comprise two central questions: 1) how do genes and environments contribute to individual differences? 2) what role does personality play? A thorough examination of both questions is not only essential for developing a comprehensive model of different modes of music engagement but could also prove to be useful for future research exploring the complex relationships between music engagement, mental health, and well-being. It could furthermore contribute to a deeper understanding of ‘musical sensibility,’ conceptualized as the disposition to be emotionally and aesthetically engaged by or feeling the music [7].

Individual differences: The role of genetic and environmental influences

To begin to answer questions about the origins of individual differences, a useful approach is to employ the classical twin design (CTD) to disentangle the relative contributions of genetic and environmental factors to trait variability. Over the past decades, a growing body of research has shown that most traits that fall within the broad definition of musicality, or the ability to perceive, produce, and enjoy music [30], are influenced by genetic effects. For example, the average heritability, i.e., the amount of trait variability that can be attributed to genetic factors [31], for objectively measured perceptual abilities (e.g., rhythm, melody, and pitch discrimination skills) and self-reported musical expertise (e.g., music aptitude or achievement) has been estimated to be around 42% [32].

Although the genetically informed literature on musicality has greatly developed in the last decades, investigations of the etiology of emotional and aesthetic musical traits have only recently emerged [7,8]. One example is a previous study reporting substantial heritability (49–64%) for self-reported ‘musical sensibility’ [7]. The affective response that a listener can experience when engaging with music, e.g., the degree of musical sensibility, is likely to be related to the goals and motivations for music engagement (i.e., the outcome, benefit or effect a listener is pursuing). However, there can be marked differences in the etiology of phenotypes that appear, on a surface level, closely related. Thus, our first aim is to provide a baseline understanding of the genetic and environmental bases of interindividual variability in the music use motivations.

Individual differences: Linking music use motivations and personality traits

Personality traits can be defined as relatively stable, individual differences in cognition, affect, and behavior [33]. In general, personality traits are often referred to as ‘broad’ when they relate to a more general set of tendencies (e.g., agreeableness) and ‘narrow’ when they pertain to more specific behaviors (e.g., compassionate, trusting, respectful). Over the past decades, there has been growing interest in associations between personality, specifically the Big Five, and various musical traits, with research focusing, for example, on their links to preferences, musical sophistication, and musical sensibility [e.g., 34–36]. Most relevant for the present context, however, is the empirical work that specifically examined the relation between individual patterns of music use and the Big Five, wherein music use most often has been assessed using the UMI [13–17]. As previously mentioned, consistent relationships have been found between open-mindedness and cognitive uses of music, with studies reporting small to moderate associations (i.e., correlations and standardized betas ranging from 0.13–0.38) [13–17]. In terms of the link between negative emotionality and musical emotion regulation, a meta-analysis [37] indicated a small-to-medium summary effect (r = 0.22, 95% confidence interval (CI) [0.17–0.27]) based on 13 correlational studies, nine of which utilized the emotion regulation subscale of the UMI. Fewer and weaker associations have typically been reported for the other three traits of the Big Five. For instance, only one study has reported a significant negative link between conscientiousness and musical emotion regulation (r = −0.22, p < 0.01) [15], whereas none of the previous studies have found support for links between agreeableness and any of the music use subscales. Results have been considerably more equivocal regarding extraversion, with studies reporting small (e.g., |r| < 0.20) positive, negative, or non-significant associations with emotional and background use of music [13,15–17].

While providing important initial insights, this previous work has key limitations, particularly concerning the psychometric and conceptual properties of the UMI, since there is little available data on its reliability and validity. Crucially, its limited conceptual bandwidth prevents a more comprehensive understanding of the associations between music use and personality traits. Finally, few of the prior studies have examined the facet-level associations with the Big Five and music use [but see, e.g., [38]. This may have masked more nuanced perspectives, as facets typically provide a richer and more precise view of criteria of interest, and there is robust evidence showing that they demonstrate modest but meaningful incremental predictive validity compared to broad traits [33,39,40], especially in terms narrow outcomes [41].

Besides these methodological shortcomings, there are also central questions that remain unanswered. The first concerns the associations with Trait Empathy – broadly defined as the dispositional tendency to understand and feel the experiences of others and which is typically considered to comprise both cognitive and emotional components [42]. The cognitive component reflects the tendency to consciously detect or understand how others think and feel, whereas the emotional component reflects the tendency to experience the perceived or assumed affective state of others. The four-dimensional Interpersonal Reactivity Index (IRI) [42,43] is a widely used self-report measure of dispositional empathy and covers the facets of fantasy, perspective taking, empathic concern, and personal distress, which tap into fictionally evoked empathy, mentalizing processes, compassion, and emotion sharing, respectively. Combined, fantasy and perspective taking are assumed to index cognitive empathy, whereas empathic concern and personal distress represent the affective aspects of empathy.

Although there is no direct empirical evidence for associations between empathy and general music use traits, global empathy has, for, example, been shown to be a robust predictor (β = 0.40, p < 0.001) of emotional responding to music [44], and the fantasy facet in particular is associated with different facets of music reward sensitivity [18]. Moreover, two empathy facets, empathic concern and fantasy, have repeatedly been linked to the paradoxical enjoyment of nominally sad music [19,20,45,46] and the closely associated phenomenon of experiencing profound feelings of being moved or touched by music [21,46]. Indeed, according to the ‘Pleasurable compassion theory’ proposed by Huron and Vuoskoski [47], individuals scoring high on empathic concern and fantasy but moderate on the personal distress facet may be motivated to engage with music because they more readily experience the pleasurable and prosocial feeling of music-induced compassion, which may be experienced as intrinsically rewarding both within and outside the musical domain.

As noted, there is considerable overlap among the Big Five and empathy facets. For example, using the IRI and two measures of the Big Five, Mooradian and Davis [22] found that empathic concern, personal distress, and fantasy were positively associated with agreeableness, negative emotionality, and open-mindedness, respectively; whereas complex and interstitial associations to the Big Five were found for perspective taking. Overall, the Big Five measures explained the greatest amount of variance in the empathic concern facet (R²= 0.36–0.44) but the least in the fantasy facet (R²= 0.12–0.19), suggesting that fantasy facet lies outside the Big Five space.

The second unanswered question relates to the nature of the relationship between music use and personality. Specifically, one possible explanation that has not yet been explored is that they may be associated because of shared genetic and/or environmental influences. Indeed, it is well-established that personality is moderately heritable, with genetic effects typically accounting for 40–50% of the variance [48–51]. Moreover, research on the relations between musical traits and other non-musical psychological characteristics (e.g., personality, intelligence) points to shared genetic influences as the main etiological factor in these associations [52–55]. Extending this line of inquiry, we recently found, using the CTD, considerable phenotypic and predominantly genetic overlap between global musical sensibility and open-mindedness (especially the aesthetic sensitivity facet), and, to a lesser extent, agreeableness (especially the compassion facet) and negative emotionality [36]. Thus, one possible account is that music use motivations and personality are linked due to common genetic and/or environmental influences.

In sum, although the existing research has succeeded in identifying key phenotypic associations between selected dimensions of music use and the Big Five, more research is needed regarding other aspects of personality and the role of shared genetic and environmental factors. Relying on well-validated and comprehensive assessments of music use motivations and personality, combining different but key personality frameworks, our second aim is to provide a detailed understanding of how music use motivations relate to different aspects of personality, both phenotypically and etiologically. Given that most previous studies have focused on domain-level associations and the previously noted advantages of using conceptually narrower personality traits, we focused here on higher-resolution associations with personality facets.

The present study

In seeking to address the two principal aims of this study, our primary objective was to provide a detailed understanding of the origins and nature of individual differences in music use motivations. To this end, we employed phenotypic- and twin analyses to examine a) the genetic and environmental basis of motivations for music use (aim 1), and b) the pattern of associations and degree of etiological overlap between the music use motivation dimensions and personality, including facets of the Big Five and empathy (aim 2).

Procedure and ethics

The data used in the present context was collected online from March 24^th through April 19^th 2022. The overall project was approved by the Regional Committee for Medical and Health Research Ethics of South-East Norway (#244965). Participants were recruited from the Norwegian Twin Registry [56] through email and/or text message invitations, which included information about the study and a link to the survey webpage. Written informed consent was obtained from all participants, and all methods were performed in accordance with relevant guidelines and regulations.

Participants

The sample for the present paper consisted of same-sex monozygotic (MZ) and dizygotic (DZ) twins born 1967–1991. In total, 2611 individual twins signed an informed consent and agreed to participate in the study (response rate = ~33%). This sample consisted of 1242 paired responders and 1369 single responders, of which 1728 were women (mean (SD) age = 43.4 (7.75), range = 31–55) and 883 were men (mean (SD) age = 44.5 (7.31), range = 31–55). Of the full twin pairs, there were 431 MZ pairs and 190 DZ pairs.

Measures

Music use motivations were measured using Module 4 of the MUSEBAQ inventory [25]. As described, it consists of four subscales, including musical transcendence (MM-transcendence; 10 items), emotion regulation (MM-emotion regulation; 9 items), social bonding (MM-social; 7 items), and music identity and expression (MM-identity; 4 items). An example item is ‘’Music raises me to another state of mind.’ Supplementary S1 Table in S1 File gives an overview of all items and descriptive statistics. Items are rated on a 5-point scale ranging from strongly disagree to strongly agree and introduced with the prompt: ‘Please tell us about your reasons for using music by ticking the circle that describes you best.’ In the current sample, Cronbach alphas ranged from 0.78 (MM-identity) to 0.93 (MM-emotion regulation).

The Big Five traits were assessed with the BFI-2 [12,57]. This 60-item inventory covers the five broad domains (extraversion, agreeableness, conscientiousness, negative emotionality, and open-mindedness) and three lower-order facets for each domain. For example, the open-mindedness domain includes the facets of intellectual curiosity, aesthetic sensitivity, and creative imagination. An example item is ‘I am someone who is complex, a deep thinker.’ All items are rated on a 5-point scale ranging from strongly disagree to strongly agree. A full overview of items can be found in Soto and John [12]. In the current sample, Cronbach alphas ranged from 0.55 (the respectfulness facet of agreeableness) to 0.82 (the sociability facet of extraversion and depression of negative emotionality).

Trait Empathy was assessed using the Brief Interpersonal Reactivity Index (B-IRI) [58]. This shortened version of the original IRI [42] consists of 16 items assessing individual differences in empathy across the facets of fantasy, perspective taking, empathic concern, and personal distress. An example item is ‘I often have tender, concerned feelings for people less fortunate than me.’ Items are rated on a 7-point scale ranging from does not describe me at all to describes me very well. A full overview of items can be found in Ingoglia et al., [58]. Cronbach alphas in the current sample ranged from 0.75 (personal distress) to 0.84 (fantasy).

Statistical analyses

All statistical analyses were performed using R (versions 4.4.1 and 4.4.2) [59] and R Studio (versions 2024.04.2.764 and 2024.9.1.394) [60]. Phenotypic correlations were calculated based on a sample of single twins and one member from each twin pair (n ranged from 1961−1987) to minimize bias. In the regression analyses, Generalized Estimating Equations (GEE) [61] and the R package geepack (version 1.3.12) [62] were used to account for the dependence of the twin data. The R packages umx (versions 4.19.00 and 4.21.0) [63] and OpenMx (versions 2.20.6 and 2.21.13) [64] with full information maximum likelihood estimation were used for all twin analyses. Model fit was determined using the minus-2-log-likelihood test (Δ − 2LL) and the Akaike’s Information Criterion (AIC) [65]. A non-significant test indicates that a nested model does not cause a significant decrease in fit relative to the full model and that the more parsimonious model is preferred, whereas a lower AIC value indicates a better fit.

Genetic and environmental influences on motivations for music use

The pattern of MZ and DZ twin correlations indicated evidence of additive genetic influences for all dimensions (Table 1). Potential C effects were indicated (i.e., rMZ < 2 x rDZ) for the MM-social subscale, whereas the DZ correlations were about half that of the MZ twin correlations for the other MM subscales. Model comparisons showed that the AE model had the best fit to the data (ΔAIC = − 8.912; p = 0.351; S6 Table in S1 File). The heritability estimates ranged from 0.38 (MM-Social) to 0.52 (MM-Identity), all of which were significant based on the 95% CIs (Table 3). In the full ACE model, however, C effects were found for the MM-Social subscale (estimated at 0.29, 95% CI [0.01; 0.53]). We therefore fitted a univariate model specifically for this subscale to determine the significance of such effects. Neither the AE model (p = 0.052) nor the CE model (p = 0.613) fit significantly worse than the full ACE model, but the CE model had the best fit according to the AIC (ΔAIC −1.74). In this model, the shared environmental variance component was estimated at 0.34, 95% CI [0.27; 0.41]. However, given the general lack of C effects previously reported for conceptually related measures [7,8] and the improved AIC results when all C components were collectively dropped from the multivariate model, we chose to proceed with the AE model in subsequent analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Estimated additive genetic- (A) and non-shared environmental (E) influences, as well as genetic (below diagonal) and environmental (above diagonal) correlations.

https://doi.org/10.1371/journal.pone.0329808.t003

As depicted in Table 3, genetic correlations between the four subscales were all significant and substantial, ranging from 0.71, 95% CI [0.63; 0.78] to 0.90, 95% CI [0.86; 0.93] (mean rA = 0.80). The unique environmental correlations were also high, ranging from 0.48, 95% CI [0.42; 0.55] to 0.75, 95% CI [0.71; 0.79] (mean rE = 0.60). Of note, the phenotypic- (r = 0.80), genetic- (rA = 0.90), and environmental (rE = 0.69) correlations between the MM-transcendental and MM-identity subscales were all substantial, indicating a high degree of shared variance among these two specific subscales.

Phenotypic associations between music use motivations and personality

Phenotypic correlations between the MM dimensions and personality facets are illustrated in Fig 1. For reference, S7 Table in S1 File gives an overview of correlations between the MM subscales and the broad personality domains (i.e., extraversion, agreeableness, conscientiousness, negative emotionality, open-mindedness, and empathy) and S8 Table in S1 File provides an overview of correlations among all primary study variables. The four MM dimensions were associated to varying degrees with the personality facets, but most correlations were positive, significant, and of small-to-medium effect size (Fig 1). The strongest correlations across all subscales were with facets related to open-mindedness followed by empathy, in particular the facets O2-aesthetic sensitivity (r range = 0.33–0.53) and IRI-fantasy (r range = 0.28–0.41). Conversely, correlations with facets of extraversion- and conscientiousness were generally low, and some were insignificant. Further, and as expected, substantial correlations were also found among and across the Big Five and empathy variables (S8 Table in S1 File).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Bivariate phenotypic correlations between the music use motivation subscales and the Big Five and empathy facets.

Significance codes (without adjusting for multiple testing): * p < 0.05; ** p < 0.01; *** p < 0.001. Facet codes (in the order they appear in the plot): BFI-E1 sociability (extraversion); BFI-E2 assertiveness (extraversion); BFI-E3 energy level (extraversion); BFI-A1 compassion (agreeableness); BFI-A2 respectfulness (agreeableness); BFI-A3 trust (agreeableness); BFI-C1 organization (conscientiousness); BFI-C2 productiveness (conscientiousness); BFI-C3 responsibility (conscientiousness); BFI-N1 anxiety (negative emotionality); BFI-N2 depression (negative emotionality); BFI-N3 emotional volatility (negative emotionality); BFI-O1 intellectual curiosity (open-mindedness); BFI-O2 aesthetic sensitivity (open-mindedness); BFI-O3 creative imagination (open-mindedness); IRI-FA fantasy (empathy); IRI-PT perspective taking (empathy); IRI-EC empathic concern (empathy); IRI-PT personal distress (empathy).

https://doi.org/10.1371/journal.pone.0329808.g001

To identify personality traits with independent effects and to inform the selection of variables for further twin modeling, we fitted four regression models (Table 4) in which all Big Five and empathy facets served as predictors of each of the four MM subscales. Overall, personality explained 37% (MM-transcendence), 28% (MM-emotion regulation), 18% (MM-social), and 29% (MM-identity) of the phenotypic variance. Of the 19 facets, seven were significant at p < 0.01 for MM-transcendence; six for MM-emotion regulation; five for MM-social; and three for MM-identity. Particularly the personality facets of O2-aesthetic sensitivity and IRI-fantasy represented consistent and robust predictors across all four MM dimensions. C3-responsibility and A3-trust were also significant predictors across all MM dimensions except MM-identity, but their influence was comparatively weak (i.e., beta’s < |0.10|). Notably, the sensitivity analysis using an O2-aesthetic sensitivity score without the item that explicitly mentions music (i.e., item 20) showed that the associations with all MM dimensions remained substantial (MM-T r = 0.48, β = 0.32; MM-E r = 0.30, β = 0.20; MM-S r = 0.34, β = 0.26; MM-I r = 0.43, β = 0.32), all highly significant (p < 0.001). Moreover, the O2-aesthetic sensitivity facet remained the strongest predictor across all MM-dimensions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Results of regression models for the associations between personality facets and music use motivations.

https://doi.org/10.1371/journal.pone.0329808.t004

In addition to these broad patterns, specific facets had significantly unique contributions according to the type of MM. That is, MM-transcendence was further positively predicted by traits such as O1-intellectual curiosity and IRI-perspective taking, while for MM-emotion regulation, positive contributions were also found from the negative emotionality facets N2-depression and N1-anxiety. Additional positive predictors of MM-social and MM-identity included, respectively, A1-compassion and E3-energy level. In contrast to IRI-fantasy, the effects of the other empathy facets (particularly empathic concern and personal distress) were largely insignificant when accounting for the Big Five, despite non-trivial and significant correlations with certain MM dimensions.

All predictors that met our original p < 0.01 threshold remained significant at p < 0.05 following FDR correction (Table 4), and as such, supports the robustness of our findings. Moreover, the specification approach also screened out other, potentially less important, predictors that were originally significant at 0.05 and which remained significant at 0.05 after corrections (e.g., the IRI-personal distress facet as a predictor of the MM-transcendence subscale). That said, it is generally not recommended to adjust alpha levels to account for multiple testing once more stringent significance thresholds have been specified [68,69].

Etiology of associations between personality facets and music use motivations

We first modeled the relationship between each of the MM dimensions and personality facets (with beta coefficients meeting the p < 0.01 significance threshold) using the Cholesky decomposition approach. Goodness-of-fit measures and results from the model comparisons summarized in S9 Table in S1 File showed that the AE model had the best fit in each case, including the MM-social subscale. An overview of the standardized parameter estimates derived from the best-fitting AE Cholesky models and the genetic and environmental correlations among all modeled variables (estimated using the direct variance approach) can be found in S10-S13 Tables in S1 File. Of note, significant genetic influences were found for all personality facets, with heritability estimates ranging from 0.24, 95% CI [0.16; 0.32] to 0.58, 95% CI [0.52; 0.63] (S14 Table in S1 File).

Fig 2 depicts the genetic and environmental variance decomposition of each MM dimension, distinguishing between personality-related influences and those unique to each dimension. Fig 3 displays the pairwise genetic and unique environmental correlations (and their 95% CIs) between the MM dimensions and personality facets included in the models.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Proportion of variance of the MM-transcendence (A), MM-emotion regulation (B), MM-social bonding (C), and MM-identity (D) subscales explained by genetic and environmental factors specific to each of them (unique genetic/environmental) or overlapping genetic/environmental factors with personality.

https://doi.org/10.1371/journal.pone.0329808.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Estimates of genetic (rA) and non-shared environmental (rE) correlations and their 95% confidence intervals between the MM dimensions and the personality variables.

Notes. Personality variables are sorted based on the magnitude of the phenotypic correlations (cf. Fig 1). Facet codes: BFI-A1 compassion (agreeableness); BFI-A3 trust (agreeableness); BFI-C3 responsibility (conscientiousness); BFI-E3 energy level (extraversion); IRI-FA fantasy (empathy); BFI-N1 anxiety (negative emotionality); BFI-N2 depression (negative emotionality); BFI-O1 intellectual curiosity (open-mindedness); BFI-O2 aesthetic sensitivity (open-mindedness); IRI-PT perspective taking (empathy).

https://doi.org/10.1371/journal.pone.0329808.g003

Seven facets were included in the MM-transcendence twin models: O2-aesthetic sensitivity, IRI-fantasy, N2-depression, A3-trust, C3-responsibility, O1-intellectual curiosity, and IRI-perspective taking. Variance decomposition showed that personality facets explained 33% and 9% of the genetic and environmental variance, respectively (Fig 2A). Of the total genetic variance in MM-transcendence (50%), the majority was explained by personality-related genetic influences (66%), whereas personality explained 17% of the total environmental variance (50%). The genetic correlations with the open-mindedness- and empathy facets were all substantial (Fig 3), ranging from 0.51, 95% CI [0.42; 0.60] (O1-intellectual curiosity) to 0.72, 95% CI [0.65; 0.79] (O2-aesthetic sensitivity). Except for IRI-perspective taking, all environmental correlations were significant and small to moderate, ranging from 0.23 for both O1-intellectual curiosity (95% CI [0.14; 0.31]) and IRI-fantasy (95% CI [0.15; 0.32]) to 0.31, 95% CI [0.23; 0.39] (O2-aesthetic sensitivity). Estimates of genetic and environmental correlations with the other facets included in the model, i.e., N2-depression, A3-trust, and C3-responsibility, were generally small (< 0.15) with large confidence intervals.

Six facets were included in the MM-emotion regulation twin models: O2-aesthetic sensitivity, N2-depression, IRI-fantasy, N1-anxiety, C3-responsibility, and A3-trust. Variance decomposition showed that personality facets explained 16% and 7%, respectively, of the genetic and environmental variance in MM-emotion regulation (Fig 2B). Of the total genetic (40%) and environmental (60%) variance, personality-related influences explained 40% and 12%, respectively. As depicted in Fig 3, positive and moderately high genetic correlations were found between MM-emotion regulation and the personality facets O2-aesthetic sensitivity (rA = 0.51, 95% CI [0.41; 0.61]), IRI-fantasy (rA = 0.44, 95% CI [0.32; 0.56]), N1-anxiety (rA = 0.30, 95% CI [0.18; 0.43]), and N2-depression (rA = 0.27, 95% CI [0.14; 0.38]), whereas environmental correlations were small and similar across these facets (rE ≈ 0.20). Both the genetic and environmental associations with A3-trust, as well as the genetic correlation with C3-responsibility, were non-significant, whereas the environmental association with C3-responsibility was significant and weakly negative (rE = −0.15, 95% CI [−0.23; −0.07]).

Five facets were included in the MM-social twin models: O2-aesthetic sensitivity, IRI-fantasy, C3-responsibility, A3-trust, and A1-compassion. Variance decomposition showed that personality facets explained 17% and 3%, respectively, of the genetic and environmental variance in MM-social (Fig 2C). Of the total genetic (38%) and environmental (62%) variance, personality-related influences explained 45% and 5%, respectively. The genetic correlations with O2-aesthetic sensitivity (rA = 0.60, 95% CI [0.50; 0.71]) and IRI-fantasy (rA = 0.44, 95% CI [0.31; 0.56]) were moderate to strong (Fig 3), in addition to small environmental correlations (rE = 0.16, 95% CI [0.07; 0.24] in both cases). The genetic correlations with the two agreeableness facets A1-compassion (rA = 0.36, 95% CI [0.20; 0.52]) and A3-trust (rA = 0.25, 95% CI [0.11; 0.40]) were small to moderate, whereas the environmental correlations were both non-significant. Neither correlation with C3-responsibility was significant.

Three facets were included in the MM-identity twin models: O2-aesthetic sensitivity, IRI-fantasy, and E3-energy level. Variance decomposition showed that personality facets explained 23% and 5%, respectively, of the genetic and environmental variance (Fig 2D). Of the total genetic (52%) and environmental (48%) variance, personality-related influences explained 44% and 10%, respectively. MM-identity had strong genetic correlations and small to moderate environmental correlations with O2-aesthetic sensitivity (rA = 0.66, 95% CI [0.59; 0.73], rE = 0.30, 95% CI [0.21; 0.38]) and IRI-fantasy (rA = 0.47, 95% CI [0.37; 0.57]; rE = 0.15, 95% CI [0.06; 0.24]), whereas the genetic and environmental associations with E3-energy level were small and equal (rA = 0.17, 95% CI [0.06; 0.28]; rE = 0.17, 95% CI [0.08; 0.25]) (Fig 3).

Although associations between MM dimensions and certain personality facets seemed to reflect trait-congruent associations, strong and consistent genetic links were found across all MM dimensions for O2-aesthetic sensitivity (rA range = 0.51–0.72, mean = 0.62) and IRI-fantasy (rA range = 0.44–0.60, mean = 0.49), in addition to small-to-moderate unique environmental correlations (O2-aesthetic sensitivity rE range = 0.16–0.31, mean = 0.24; IRI-fantasy rE range = 0.15–0.23, mean = 0.18). The two facets also shared substantial etiological variance with each other (rA and rE about 0.57, 95% CI [0.48; 0.65] and 0.19, 95% CI [0.10; 0.27], respectively; see S10-S13 Tables in S1 File). Given the complexity of the etiological relationships between personality variables, a thorough examination of their unique contribution to the genetic and environmental covariance with MM dimensions is beyond the scope of this study. However, as can be seen in S10-S13 Tables in S1 File, IRI-fantasy had significant unique genetic covariance, over and above O2-aesthetic sensitivity, for two of the four MM dimensions (MM-transcendence and MM-emotion regulation), in addition to unique environmental covariance with all MM dimensions. Importantly, these findings suggest that the associations between IRI-fantasy and MM dimensions are not solely explained by its shared variance with O2-aesthetic sensitivity, mirroring results from the regression analyses.

Etiology of different dimensions of music use motivation

Our first aim was to determine the relative importance of genetic and environmental factors in shaping individual differences. Heritability estimates were 50%, 40%, 38%, and 52% for MM-transcendence, MM-emotion regulation, MM-social, and MM-identity, respectively, with an average heritability of 45%. Individual-specific environmental influences accounted for the remaining variance. These estimates are consistent with the broader genetically informed literature on musicality [32], as well as the heritability of the five music reward facets (range = 42%-52%, mean = 47%) [8], but they appear to be somewhat lower than what we previously found for the four musical sensibility facets (range = 49%−65%, mean = 56%) and general musical sensibility (64%) [7]. Thus, these initial findings suggest that compared to musical sensibility, individual differences in the various dimensions of MM are more influenced by environmentally shaped unique experiences than by genetic factors.

Although the multivariate modeling results indicated significant additive genetic and non-shared environmental influences, it is worth noting that, in the univariate model for the MM-social subscale specifically, shared environment influences were significant, whereas additive genetic influences were not. This latter finding contrasts with the earlier research on the etiology of musical sensibility [7] and music reward sensitivity [8]. In fact, in the latter study, ‘Social reward’ had the highest heritability (52%) among the five music reward facets and the authors reported no indications of shared environmental influences.

In the present results, the lack of significant genetic effects (in the univariate model) likely reflects limited power to distinguish between familial effects, especially when one of the components (A or C) is small. Moreover, multivariate models, with more information, generally involve greater power and, consequently, more robust estimates than univariate models. Nevertheless, our results do indicate that the heritability of this subscale is likely modest and, along with its genetic correlations with other variables, should thus be interpreted cautiously. This further suggests that the genetic and environmental architecture may differ across the four subscales and that there might be something special about the MM-social subscale, which should be more comprehensively investigated in future studies using alternative modeling approaches.

Another noteworthy finding was the considerably high associations among the MM subscales: phenotypically (r range = 0.59–0.80), genetically (rA range = 0.71–0.90), and environmentally (rE range = 0.48–0.75). Importantly, high inter-factor correlations were also reported in the original publication by Chin et al. [25] (range 0.75–0.90), suggesting that these findings are not just due to chance or specific to the present sample. These findings can be interpreted in different ways, one of which relates to questions about the distinctiveness of the four subscales. In particular, the especially strong associations between the MM-transcendence and MM-identity subscales may indicate that the two subscales reflect highly overlapping concepts. This would accord with the empirical findings by Schäfer et al. [1], wherein the first of the three identified dimensions (self-awareness) encompassed items indexing functions related to, e.g., self-related thoughts, identity, meaning, absorption, and escapism [cf. Table A3 in Schäfer et al. 1], echoing the conceptual content of the MM-transcendence and MM-identity dimensions. However, our preliminary CFA results, combined with the fact that none of the CIs around the phenotypic correlations reached 1, provide tentative support for the notion that, while related, the four subscales do not measure the same thing, supporting a multidimensional structure.

Another interpretation is that the overall high associations between dimensions reflect the presence of a common underlying latent factor, or alternative data-generating mechanisms, like those postulated by network models [72]. Since we did not comprehensively assess the underlying latent structure, the present findings strongly warrant further examination of the structure and dimensionality of the MM scale.

Links between music use motivations and personality

Our second aim was to examine the role of personality traits (the Big Five and Trait Empathy) in individual differences in motivations for music use. To this end, we performed a comprehensive assessment of facet-level associations across MM dimensions at the phenotypic and etiological levels.

Limitations

The work presented here should be interpreted in the context of certain limitations. First, the usual limitations of the CTD apply, including limited power to distinguish between different genetic mechanisms (i.e., additive and non-additive genetic effects) and familial influences (i.e., additive genetic and shared environmental influences) [82–84]. Indeed, the latter issue was raised for the MM-social subscale. Therefore, the present results do not rule out the presence of possible dominance genetic and shared environmental effects and the final AE models should thus be regarded as tentative. Moreover, the obtained estimates are limited to the specific sample (in terms of age, socioeconomic status, culture, and ethnicity) and the measurement that was used. The present sample mostly comprised middle-aged women from a rather culturally homogenous group. Although we controlled for sex and age effects, future studies should seek to examine sex-specific patterns and replicate the reported estimates in culturally more diverse samples and across different age groups, ideally using longitudinal designs. Indeed, previous studies points to age differences in observed music engagement styles [85], as well as associations with personality [86], in addition to cultural differences in how music is used to achieve various well-being goals [87]. For example, during the COVID-19 crisis, music was more prominently used as a resource for fostering a sense of togetherness in collectivistic cultures compared to individualistic ones [87]. Future work should also include additional measures of MM (e.g., behavioral tasks, physiological recordings) to complement self-reports, as well as alternative twin modeling approaches.

Second, it is important to highlight that genetic correlations do not imply causation, nor do they inform the directionality of the causal pathways – i.e., the genetic influence on which measure impacts the other, cannot be established in our correlational data. Third, all measures were assessed using self-reports, which might introduce shared method variance. If present, this may have artificially inflated the phenotypic- and non-shared environmental correlations but not genetic correlations. Fourth, in terms of the more modest and seemingly less congruent phenotypic associations between certain MM dimensions and facets of, e.g., conscientiousness and agreeableness, it is important to note that most of the genetic and environmental correlations underpinning these relationships were generally low and imprecise, and in some cases, both estimates were insignificant (e.g., in the link between the trust facet of agreeableness and the MM-emotion regulation dimension). As such, these relationships should be cautiously interpreted, as they may be spurious findings. Finally, while we used a comprehensive MM measure, it is likely not exhaustive. For example, it could be argued that the lack of a dimension capturing ‘background’ use might obscure meaningful relations to certain personality traits. Indeed, for many, the use of music may serve more utilitarian purposes, such as enhancing concentration, providing physical stimulation during exercise, offering entertainment, or creating background noise to eliminate silence. Relatedly, it should also be mentioned that there is a lack of items that capture the upregulation of positive emotions (in the emotion regulation subscale), such as using music to experience joy. That said, research suggests that negative emotions are most often the target of regulation in everyday life [88].